Add spec parser

Currently, the parser only supports definitions of lexical specification.

1 files changed, 137 insertions, 0 deletions

diff --git a/spec/lexer.go b/spec/lexer.go
new file mode 100644
index 0000000..54eb13a
--- /dev/null
+++ b/spec/lexer.go
@@ -0,0 +1,137 @@
+//go:generate maleeni compile -l lexspec.json -o clexspec.json
+
+package spec
+
+import (
+	_ "embed"
+	"encoding/json"
+	"fmt"
+	"io"
+	"strings"
+
+	mldriver "github.com/nihei9/maleeni/driver"
+	mlspec "github.com/nihei9/maleeni/spec"
+)
+
+type tokenKind string
+
+const (
+	tokenKindID              = tokenKind("id")
+	tokenKindTerminalPattern = tokenKind("terminal pattern")
+	tokenKindColon           = tokenKind(":")
+	tokenKindSemicolon       = tokenKind(";")
+	tokenKindEOF             = tokenKind("eof")
+	tokenKindInvalid         = tokenKind("invalid")
+)
+
+type token struct {
+	kind tokenKind
+	text string
+}
+
+func newSymbolToken(kind tokenKind) *token {
+	return &token{
+		kind: kind,
+	}
+}
+
+func newIDToken(text string) *token {
+	return &token{
+		kind: tokenKindID,
+		text: text,
+	}
+}
+
+func newTerminalPatternToken(text string) *token {
+	return &token{
+		kind: tokenKindTerminalPattern,
+		text: text,
+	}
+}
+
+func newEOFToken() *token {
+	return &token{
+		kind: tokenKindEOF,
+	}
+}
+
+func newInvalidToken(text string) *token {
+	return &token{
+		kind: tokenKindInvalid,
+		text: text,
+	}
+}
+
+type lexer struct {
+	s      *mlspec.CompiledLexSpec
+	d      *mldriver.Lexer
+	dufTok *token
+}
+
+//go:embed clexspec.json
+var lexspec []byte
+
+func newLexer(src io.Reader) (*lexer, error) {
+	s := &mlspec.CompiledLexSpec{}
+	err := json.Unmarshal(lexspec, s)
+	if err != nil {
+		return nil, err
+	}
+	d, err := mldriver.NewLexer(s, src)
+	if err != nil {
+		return nil, err
+	}
+	return &lexer{
+		s: s,
+		d: d,
+	}, nil
+}
+
+func (l *lexer) next() (*token, error) {
+	for {
+		tok, err := l.d.Next()
+		if err != nil {
+			return nil, err
+		}
+		if tok.Invalid {
+			newInvalidToken(tok.Text())
+		}
+		if tok.EOF {
+			return newEOFToken(), nil
+		}
+		switch tok.KindName {
+		case "white_space":
+			continue
+		case "newline":
+			continue
+		case "identifier":
+			return newIDToken(tok.Text()), nil
+		case "terminal_open":
+			var b strings.Builder
+			for {
+				tok, err := l.d.Next()
+				if err != nil {
+					return nil, err
+				}
+				if tok.EOF {
+					return nil, synErrUnclosedTerminal
+				}
+				switch tok.KindName {
+				case "pattern":
+					// Remove '\' character.
+					fmt.Fprintf(&b, strings.ReplaceAll(tok.Text(), `\"`, `"`))
+				case "escape_symbol":
+					return nil, synErrIncompletedEscSeq
+				case "terminal_close":
+					return newTerminalPatternToken(b.String()), nil
+				}
+			}
+		case "colon":
+			return newSymbolToken(tokenKindColon), nil
+		case "semicolon":
+			return newSymbolToken(tokenKindSemicolon), nil
+		default:
+			return newInvalidToken(tok.Text()), nil
+		}
+	}
+}